With the expansion and the diversification of current networks in terms of applications, technologies and services, the network control and management are becoming overly hard to manage manually even for skilled administrators. The autonomic computing and the knowledge plane are two promising initiatives proposed to cope with the ever-growth network complexity. In this article, we propose to use the piloting plane concept to grant autonomicity in current networks. This plane aims at optimizing existing applications by using elaborate algorithms fed by pertinent knowledge recovered from the knowledge plane. In our approach, we demonstrate that the knowledge and the piloting planes are complementary and both are required when implementing autonomic solutions. We illustrate our proposal by applying these concepts on the handover selection phase which is a key issue in the handover process for granting global mobility to users in heterogeneous environment. We also describe a first implementation of our handover piloting system. Results from preliminary tests in real environments scenario exhibit the effectiveness of our system.